Installation#

Python version 3.9 or newer is recommended. Warp can run on x86-64 and ARMv8 CPUs on Windows, Linux, and macOS. GPU support requires a CUDA-capable NVIDIA GPU and driver (minimum GeForce GTX 9xx).

The easiest way to install Warp is from PyPI:

$ pip install warp-lang

Pre-built binary packages are also available on the Releases page. To install in your local Python environment extract the archive and run the following command from the root directory:

$ pip install .

Dependencies#

Warp supports Python versions 3.7 onwards, with 3.9 or newer recommended for full functionality. Note that some optional dependencies may not support the latest version of Python.

NumPy must be installed.

The following optional dependencies are required to support certain features:

usd-core: Required for some Warp examples, warp.sim.parse_usd(), and warp.render.UsdRenderer.
JAX: Required for JAX interoperability (see JAX).
PyTorch: Required for PyTorch interoperability (see PyTorch).
NVTX for Python: Required to use wp.ScopedTimer(use_nvtx=True).

Building the Warp documentation requires:

Building from source#

For developers who want to build the library themselves the following tools are required:

Microsoft Visual Studio (Windows), minimum version 2019
GCC (Linux), minimum version 9.4
CUDA Toolkit, minimum version 11.5
Git Large File Storage

After cloning the repository, users should run:

$ python build_lib.py

This will generate the warp.dll / warp.so core library respectively. It will search for the CUDA Toolkit in the default install directory. This path can be overridden by setting the CUDA_PATH environment variable. Alternatively, the path to the CUDA Toolkit can be passed to the build command as --cuda_path="...". After building, the Warp package should be installed using:

$ pip install -e .

Which ensures that subsequent modifications to the library will be reflected in the Python package.

Conda environments#

Some modules, such as usd-core, don’t support the latest Python version. To manage running Warp and other projects on different Python versions one can make use of an environment management system such as Conda.

WARNING: When building and running Warp in a different environment, make sure the build environment has the same C++ runtime library version, or an older one, than the execution environment. Otherwise Warp’s shared libraries may end up looking for a newer runtime library version than the one available in the execution environment. For example on Linux this error could occur:

OSError: <...>/libstdc++.so.6: version `GLIBCXX_3.4.30' not found (required by <...>/warp/warp/bin/warp.so)

This can be solved by installing a newer C++ runtime version in the runtime conda environment using conda install -c conda-forge libstdcxx-ng=12.1 or newer. Or, the build environment’s C++ toolchain can be downgraded using conda install -c conda-forge libstdcxx-ng=8.5. Or, one can activate or deactivate conda environments as needed for building vs. running Warp.